PERFORMANCE EVALUATION OF ADVANCED CONTROL ALGORITHMS ON A FOPDT MODEL C I KASTAMONITIS, G P SYRCOS & A DOUNIS Automation Department Technological Educational Institute of Piraeus Petrou Ralli & Thivon GREECE CKastam@yahoo.com Abstract: This paper presents the simulation of a simple First Order plus Delay Time (FOPDT) process model using advanced control algorithms Specifically, these advanced algorithms are the IMC-based PID controller, the Model Predictive Controller (MPC) and the Proportional-Integral-Plus Controller (PIP) and their performance is compared with the conventional Proportional – Integral – Derivative (PID) algorithm The simulations took place using the Matlab/Simulink™ software Key-Words: FOPDT, advanced control algorithms, IMC-based PID, MPC, PIP, PID, Matlab/Simulink™ Introduction To date, the most popular control algorithm used in industry is the ubiquitous PID controller which has been implemented successfully in various technical fields However, since the evolution of computers and mainly during the 1980s a number of modern and advanced control algorithms have been also developed and applied in a wide range of industrial and chemical applications Some of them are the Internal Model – based PID controller, the Model Predictive controller and the Proportional-Integral-Plus controller The common characteristic of the above algorithms is the presence in the controller structure an estimation of the process’ model The purpose of this paper is to apply these advanced algorithms to a linear first order plus delay time (FOPDT) process model and compare their step response with the conventional PID controller Initially, it will be presented a brief discussion over the theoretical designing aspects of each applied algorithm The main section of the paper is devoted to the simulation results in terms of type servomechanism performance of a simple FOPDT process, using the above control algorithms in various practical scenarios 1.1 Proportional – Integral – Derivative Controller The Proportional – Integral – Derivative (PID) control algorithm is the most common feedback controller in industrial processes It has been successfully implemented for over 50 years, as it provides satisfactory robust performance despite the varied dynamic characteristics of a process plant [1] The proper tuning of the PID controller aims a desired behavior and performance for the controlled system and refers to the proper definition of the parameters which characterize each term Over the past, it has been proposed several tuning methods, but the most popular (due to its simplicity) is the ZieglerNichols tuning method This tuning method is based on the computation of a process’s critical characteristics, i.e critical gain Kcr and critical period Pcr [2] Table summarizes the computation of PID parameters [3] Controller P KP K cr TI PI K cr 2 Pcr 1.2 PID K cr 1.7 Pcr TD Pcr Table 1: Ziegler-Nichols PID tuning computation 1.2 IMC-based PID Controller The internal model control (IMC) algorithm is based on the fact that an accurate model of the process can lead to the design of a robust controller both in terms of stability and performance [4] The basic IMC structure is shown in Figure and the controller representation for a step perturbation is described by (1) G f ( s) Gq ( s ) (1) G ( s) mm where Gmm (s ) is the inverse minimum phase part of the process model and G f (s) is a nth order low pass filter ( λs 1) n The filter’s order is selected so that Gq (s ) is semi-proper and λ is a tuning parameter that affects the speed of the closed loop system and its robustness [7] Figure 1: IMC control structure However, there is equivalence between the classical feedback and the IMC control structure, allowing the transformation of an IMC controller to the form of the well-known PID algorithm Gq ( s ) Gc ( s ) (2) G ( s )G ( s ) m variables in order to optimize the future behavior of the controlled process Initially, it has been developed to accomplish the specialized control needs in power plants and oil refineries However because its ability to handle easily constraints and MIMO systems with transport lag, it can be used in various industrial fields [8] The first predictive control algorithm is referred to the publication of Richalet et al titled “Model Predictive Heuristic Control” [9] However, in 1979, Cutler and Ramaker by Shell™ developed their own MPC algorithm named Dynamic Matrix Control – DMC [10] Since then, a great variety of algorithms based on the MPC principle has been also developed Their main difference is focused on the use of various plant models which is an important element of the computation of the predictive algorithm (i.e step model, impulse model, state-space models, etc) Figure shows a typical MPC block diagram q The resulted controller is called IMC-based PID controller and has the usual PID form (3) Gc ( s ) K p 1 TD s (3) T Is IMC-based PID tuning advantage is the estimation of a single parameter λ instead of two (concerning the IMC-based PI controller) or three (concerning the IMC-based PID controller) The PID parameters are then computed based on that parameter [4] Though for the case of a FOPDT (4) process model, the delay time should be approximated first by a zero-order Padé (usually) approximation [6] However, the IMC-based PID tuning method can be summarized according to the following Table [7] k G s c e θs (4) τs Figure 2: MPC block diagram The main idea of the predictive control theory is derived from the exploitation of an internal model of the actual plant, which is used to predict the future behavior of the control system over a finite time period called prediction horizon p (Figure 3) This basic control strategy of predictive control is referred to as receding horizon strategy [11] Controller K P K c TI TD λθ IMC-based τ θ τ >1.7 PI without λ Padé 2τ θ θ IMC-based τ >1.7 PI 2λ 2τ θ τθ θ IMC-based τ >0.8 PID 2λ θ 2τ θ Table 2: IMC-based PID tuning parameters of a FOPDT process 1.3 Model Predictive Controller MPC refers to a class of advanced control algorithms that compute a sequence of manipulated Figure 3: Receding Horizon Strategy Its main purpose is the calculation of a controlled output sequence y(k) that tracks optimally a reference trajectory y0(k) during m present and future control moves (m ≤ p) Though m control moves are calculated at each sampled step, only the first Δû(k)=(u0(k)-u(k)) is implemented At the next sampling interval, new values of the measured output are obtained Then the control horizon is shifted forward by one step and the above computations are repeated over the prediction horizon In order to calculate the optimal controlled output sequence, it is used a cost function of the following form [12] p J Γly [ y (k l | k ) y (k l )] l 1 m Γlu (4) Δuˆ (k l 1 l 1 where Γly and Γlu are weighting matrices used to penalize particular components of output and input signals respectively, at certain future intervals The solution of the LQR control problem is resulted to a feedback proportional controller estimated as the gain matrix k solution of the wellknown Riccati equation over the prediction horizon (5) u ( k ) kxk 1.4 PIP Controller PIP controller comprises a part of the True Digital Control – TDC control method and can be considered as a logical extension to the conventional PI/PID controller but with inherent model predictive control action The power of the PIP design derives from its exploitation of a specialized Non-Minimal State Space (NMSS) representation of a linear and discrete system referred as NMSS/PIP formulation [13] [14] The fact that the PIP is considered as a logical extension of the conventional PI/PID controlled can be appeared better when the process’s transfer function is second order of higher or includes transport lag greater than one sampling interval Then PIP controller includes also a dynamic feedback and input compensation introduced “automatically” by the specialized NMSS formulation of the control problem [15] that in general, has a numerous advantages against other advanced control structures [16] Any linear discrete time and deterministic SISO ARIMAX model can be represented by the following specialized NMSS equations (6) x (k ) Fx(k 1) qu (k 1) dy d (k ) y (k ) hx(k ) (7) q where the vectors F , , d and h comprise the parameters of the above equations [14] In the specialised NMSS/PIP case, the nonminimum n+m state vector x(k) consists not only in terms of the present and past sampled value of the output variable y(k) and the past sampled values of the input variable u(k) (as it happens in the conventional NMSS design) but also of the integral-of error state vector z(k) introduced to ensure Type servomechanism performance, i.e x(k ) y (k ), y (k 1), , y (k n 1), (8) u (k 1), , u (k m 1), z (k ) T The integral-of error state vector z (k ) defines the difference between the reference input (setpoint) y0 ( k ) and the sampled output y (k ) z ( k ) z ( k 1) { y d ( k ) y ( k )} (9) The control law associated with the NMSS model results to the usual State –Variable Feedback (SVF) form u (k ) kx( k ) (10) where k is the n m SVF gain vector The control gain vector may be easily calculated by means of a standard LQ cost function J x(i )T Qx(i ) Ru (i ) (11) i 0 where Q is a n m n m weighting matrix and R is a scalar input u (i ) weight It is worth noting that, because of the special structure of the state vector x(k), the weighting matrix Q is defined by its diagonal elements, which are directly associated with the measured variables and integral-of error state vector For example the diagonal matrix can be defined in the following default form Q diag q1 K qn 1qy213n qn 1 K qn m 1 qn m 14243 { qe qu 1 m (12) The SVF gains are obtained by the steady-state solution of the well-known discrete time matrix Riccati Equation [17], given the NMSS system description (F and q vectors) and the weighting matrices (Q and R) k f f1 f n g1 g m k1 (13) In a conventional feedback structure, the SVF controller can be implemented as shown in Figure 4, where it becomes clear how the PIP can be considered as a logical extension of the conventional PI/PID algorithm, enhanced by a higher-order forward path and input/output feedback compensators 1 1 G [ z , m 1] and F [ z , n 1] respectively [15] (14) F ( z ) f f1z f n 1z ( n 1) G ( z ) 1 g1 z g m z ( m 1) (15) weight Γly 0 , control horizon m 10 and infinity prediction horizon Whether the absence of measured disturbances or not, the ‘default’ LQ weight matrices for the PIP controller are; Qdiag 1 0.25 0.25 0.25 1 , R 0.25 , (absence of measured disturbances) and Qdiag 1 0.25 0.25 0.25 0 , R 0.25 , (presence of measured disturbances) Problem Solution Figure 4: PIP feedback block diagram Problem Formulation In order to asses the practical utility of the above described advanced control algorithms, a series of implementation simulations have been conducted on a simple FOPDT process For comparison purposes, a conventional PID controller is also designed using the Ziegler-Nichols method The FOPDT process model is described by (16) and initially is assumed absence of plant model mismatch, inputs constraints or measured disturbances The model selection is based on the fact that a FOPDT model represents any typical SISO chemical process The simulation took place using the Matlab/Simulink ™ software and the results are discussed in terms of Type servomechanism performance 0.3s G(s) e (16) s 1 The next simulation scenario includes constraints in the input manipulated variables 2 u (t ) (17) In the final simulation scenario a simple disturbance model described by (18) is also implemented, in order to study the capability of each controller in disturbance rejection 0.8 0.1s (18) Gd ( s ) e s 1 The critical characteristics for the estimation of PID parameters (See Table 1) are Kcr=5.64 and Pcr=1.083 The IMC-based PID parameters are estimated according to Table selecting 0.5 and n The calculation of MPC gain matrix includes the following parameters; input weight Γlu 1 , output With no disturbances and input constraints, the output response (Figure 6) for the advanced control algorithms yields satisfactory step behavior with good set point tracking and smooth steady state approach However, the response of the conventional PID seems to be rather disappointing, as it yields a large overshoot Figure demonstrates their control action response Mainly concerning MPC and PID algorithms, the initial sharp increase of their control action signal may not be acceptable during a practical realization of the controller in an actual industrial plant Figure shows the output response after the introduction of input constraints defined by (17) According to the results, both PIP and IMC-based PID controllers were unaffected by the input constraints as their constrained control action response has been within the constrained limits Although the response of the conventional PID controller retained its large overshoot, the introduction of input constraints has optimized its smoothness Finally MPC maintained its satisfactory performance, although the fact that its manipulated variable has been constrained the most (Figure 9) Figure 10 demonstrates the output responses of the process during the introduction of measured disturbances defined by (18) According to the results, MPC controller yields the most optimal response while PIP controller sustains its performance On the contrary IMC-based PID as well as the conventional PID yield a rather large overshoot Table shows an approximate numerical evaluation of the control algorithms for each scenario The evaluation parameters are the Overshoot (O), Rise Time (RT), Settling Time (ST), Integral Square Error (ISE), Robust stability (RS) and Robust Performance (RP) Controller Scenario PID %O 49.8 IMC-based PID 1.76 MPC 0.00 PIP 0.00 RT ST ISE 0.530 1.1800 0.0021 1.250 1.930 1.3800 0.0021 1.450 0.49 Scenario PID 50.0 0.970 0 IMC-based PID 2.00 1.240 MPC 0.00 0.8500 PIP 0.00 1.250 0.51 ??? 0.65 2.6600 0.67 Figure 7: Unconstrained Control Action step response 1.440 0.950 1.450 0.52 ??? 0.65 Scenario PID 62.9 IMC-based PID 16.2 MPC 0.00 PIP 7.38 0.530 1.930 0.48 0 0.7800 3.380 0.40 0.0021 0.0021 ??? 0.950 1.950 0.50 0 Table Numerical Evaluation of Control Algorithms Figure 7: Output Step Response with Input Constraints Figure 6: Unconstrained Output Step Response Figure 9: Constrained Control Action Step Response Figure 10: Output Step Response with Measured Disturbances [2] Ziegler, J., C., Nichols, N., B., Optimum Settings for Automatic Controllers Trans A.S.M.E vol 64., 1942 [3] Astrom K., J., & Haugglund T PID Controllers: Theory, Design and Tuning Instrument Society of America, 1995 [4] Coughanour, D Process Systems Analysis & Control NY, McGraw-Hill c1991 Chemical Engineering Series, 1991 [5] Rivera, E., D., Internal Model Control: A Comprehensive View Arizona State University, Arizona, 1999 [6] Bequette, B W., Process Control: Modelling, Design and Simulation Prentice Hall, Upper Saddle River, NJ, 2003 [7] Morari, M., & Zafiriou, F., Robust Process Control Prentice Hall, Englewood Cliffs NJ, 1989 Naeem, W., Model Predictive Control of an Autonomous Underwater Vehicle Department of Mechanical and Marine Engineering, The University of Plymouth, 2003 Conclusion This paper discusses the effect of three advanced control algorithms on a FOPDT process model in terms of type servomechanism performance These algorithms are the IMC-based PID controller, the Model Predictive controller and the PIP controller After their implementation in the FOPDT process their step response was simulated using the Matlab/Simulink™ software and compared with the conventional PID controller in various practical scenarios Such scenarios include the implementation of input constraints or measured disturbances According to the simulations results, all the advanced control algorithms perform satisfactory step behavior with good set point tracking and smooth steady state approach They also sustain their robustness and performance during the introduction of input constraints or measured disturbances Surprisingly, the step response of the conventional PID controller wasn’t as optimal as it has been expected as its overshoot exceeds any typical specification limits Acknowledgements Authors would like to thank Ioannis Sarras for the provision of useful papers concerning the PIP theory References [1] Willis, M., J., Proportional – Integral – Derivative Control Dept of Chemical and Process Engineering, University of Newcastle, 1999 [8] [9] Richalet, J., Rault, A., L.Testud, J., and Papon, J Model Predictive Heuristic Control Automatica, 14:413, 1978 [10] Cutler, C R., Dynamic Matrix Control: An Optimal Multivariable Control Algorithm With Constraints PHD thesis, University of Houston, 1983 [11] Maciejowski, J M., Predictive Control with Constraints Addison-Wesley Pub co, edition, 2001 [12] Morari, M., Ricker, N., Model Predictive Control Toolbox for Use with Matlab User’s Guide, The Mathworks Inc, 1998 [13] Hesketh, T., State-space pole-placing self-tuning regulator using input-output values IEEE Proceedings Part D, 129, 123±128, 1982 [14] Young, P.C., Behzadi, M.A., Wang, C.L., and Chotai, A., Direct digital and adaptive control by input-output, state variable feedback pole assignment International Journal of Control, 46, 1867± 1881, 1987 [15] Taylor C., J., Leigh, P., Price, L., Young, P., C., Vranken, E., Berckmans, D., ProportionalIntegral-Plus (PIP) Control of Ventilation Buildings Pergamon Control Engineering Practice, 2003 [16] Taylor, C., J., Chotai, A., Young, P., C., State space control system design based on nonminimal state-variable feedback: further generalization and unification results International Journal of Control, Vol 73, No 14, 1329-1345, 2000 [17] Astrom, K., J., & Wittenmark, B., Computer controlled systems: Theory and Design Prentice-Hall Information and System Sciences Series, 1984 ... asses the practical utility of the above described advanced control algorithms, a series of implementation simulations have been conducted on a simple FOPDT process For comparison purposes, a. .. yields a large overshoot Figure demonstrates their control action response Mainly concerning MPC and PID algorithms, the initial sharp increase of their control action signal may not be acceptable... comprises a part of the True Digital Control – TDC control method and can be considered as a logical extension to the conventional PI/PID controller but with inherent model predictive control action